Effects in Tellurium. 61 



The difference of potential between the copper leads connected 

 to opposite edges of the plate is the resultant of the Nernst 

 effect and the thermo-electric effect at the two junctions. 

 Since the two junctions are in general at slightly different 

 temperatures, they do not balance one another. If their tem- 

 peratures were constant, the thermo-electric effect could be 

 eliminated by taking readings for both directions of the mag- 

 netic field ; for in one case the Nernst effect would be added 

 to it and in the other case subtracted. Owing to the Leduc 

 effect, the temperatures do not remain constant, but are 

 altered by reversing the field. 



Let E=Nernst effect; V and Y^the observed P.D.'s for 

 two directions of the field; e— thermo-electric power for cop- 

 per and tellurium ; ^ and tf 3 =the temperatures of the two 

 junctions for one direction of the field, t/ and t 2 ' for the other ; 



T=Leduc effect=-J &-*/+£/-£/)• Then 

 V=e(f 1 -* 2 )+E 



V'=«s(* x '-*,')r-E 



whence 



Y— V 

 E=- --eT. 



2 



It is thus seen that the observed values of Y must be cor- 

 rected for the Leduc effect. In bismuth this correction was 

 found negligible ; in tellurium it is not so, owing to the enor- 

 mous thermo-electric power of copper and tellurium, but 

 assumes a magnitude of the same order as the E"ernst effect. 

 An accurate determination of the Nernst effect requires an 

 accurate knowledge not only of one value of this thermo- 

 electric power, but also of its variation with temperature and 

 with strength of field. The values of e observed are slightly 

 over 600 microvolts, whereas the value usually given is about 

 500. 



Observations of the Nernst effect give a value of Q = 0'36 at 

 33° in a field of about 3,000 gausses. As in the case of the 

 Hall effect the constant is larger than has been observed in any 

 metal. 



Change of heat-conductivity. — The conductivity is lessened 

 in the magnetic field. The relative conductivities were calcu- 

 lated from observations of temperature at four points along 

 the center line of the plate, assuming the thickness to be uni- 

 form and the surface-conductivity to be constant. Neither of 

 these conditions is fulfilled, and hence the results are only 

 approximate. The ratio of conductivity in a field of strength 

 4700 gausses to that in zero field was found to be 0*9. 



Longitudinal thermo-magnetic effect. — The warm end of the 



